I've heard the sentiment that "personal data is highly radioactive", in that collecting it at all has unforeseen consequences.
a famous example from history includes census data on Jewish people in the Netherlands being used in the Holocaust; and how significantly fewer French Jews died because France simply didn't collect that data.
but there's a specific example I recall, about a more subtle element of this. I can't find the original source for this; searching for a good half an hour yielded nothing last I tried. but the story is about how even seemingly innocuous and necessary data can be used to draw correlations and deduce much more sensitive attributes.
the details are a bit fuzzy, but paraphrasing, it goes something like this:
there was a power company, who wanted to charge people more accurately for their utility use. previously, I guess they were just playing a flat rate? not sure. so, the power company went out to every home and installed an electricity meter. to measure the power usage in that home. they collected this data, for the intent of calculating the bills and probably also to better anticipate loads on the grid and plan more efficiently. the power company now has statistics about the power usage of any given home over time. surely nothing could go wrong here.
during the holy month of Ramadan, Islamic homes fast during the day. again, the details on this are a bit fuzzy, but i understand kitchen appliances to be very power hungry, so it seems plausible to me that this directly translated to abnormal power usage. perhaps significantly lower during the day, and significantly higher after sunset.
and an employee at the power company notices this anomaly in the data. "why are so many homes acting unexpectedly for 30 days a year?". and they realize that this is Ramadan. and they realize that. "oh shit. we've accidentally created a very accurate dataset of every muslim home in this region. oh fuck."
(I'm not sure what the deal was with installing the meters? maybe they had meters already but upgraded them to "smart" ones that phone home? again, for the legit purpose of better anticipating grid load or whatever, this was not an unnecessary thing to collect, and yet, the dataset encodes a highly sensitive attributes that you absolutely do not want to collect)
what power company was this? whose words am I paraphrasing? this story didn't just like, spawn suddenly in my mind palace, did it? I strongly recall the phrase "personal data is radioactive"/"information is radioactive" or some variation on those being used in the source I've heard it from. specifically the adjective "radioactive" I'm very confident about.
Mastodon Feed
